Google Cloud deepened its push into so-called “agentic” artificial intelligence by rolling out six specialised AI agents designed to automate much of the grunt work handled by enterprise data teams. The tools, embedded across BigQuery, Spanner and related services, can build data pipelines, migrate databases, run exploratory analysis and even write Python code through natural-language prompts. Google says the agents attack the industry’s “80 per-cent toil” problem, promising faster insights while keeping data governance controls intact. Canadian start-up Cohere followed with the formal launch of North, a platform that lets large language-model agents run inside customers’ own infrastructure—ranging from private clouds to air-gapped GPUs—so that sensitive data never leaves corporate firewalls. The company, which has attracted US$970 million in funding at a US$5.5 billion valuation, said early pilots at Royal Bank of Canada, Dell and LG show the system can draft documents, summarise reports and generate market research while meeting GDPR, SOC-2 and ISO 27001 standards. Investors and security vendors are also moving quickly. Tavily raised US$25 million in Series A financing to give enterprise agents compliant access to web data, while cyber-firms such as Tenable and Straiker introduced products aimed at identifying the new risks that autonomous AI systems pose. Palo Alto Networks cautioned that threat actors are already leveraging generative AI to compress attack cycles from days to minutes, underscoring why the fast-growing agent market is pairing new productivity gains with heightened security demands.